What is ML Deployment?
The 4 Deployment Paradigms
Data Science != Data Engineering
Data science is scientific
Data engineers are concerned with
DevOps = software development + IT operations
ModelOps = data modeling + deployment operations
All the (DevOps) things!
And then more things!
ML pipeline w/ featurization logic
Monitoring + Alerting
CI/CD pipeline for automation
Testing framework (unit + integration)
Data and model drift
Reproducibility: data, code, environment, debugging
Standards for each deployment paradigm
Managed by an admin
Clear responsibilities on maintenance in production
Who gets paged at 2 in the morning?
Quantization: reduce precision of mathematical operations
Weight pruning: reduce size of architecture
Model topography: retrain using different architectures
Apply the same logic to training and scoring data
Look into MLflow’s pyfunc
Confirm production data is available in training